debugging-and-error-recovery
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides standard procedural guidance for software troubleshooting and maintenance without introducing malicious patterns or dangerous operations.
- [INDIRECT_PROMPT_INJECTION]: The skill explicitly mitigates indirect prompt injection risks in the section 'Treating Error Output as Untrusted Data'. It identifies error outputs, stack traces, and logs as untrusted ingestion points and instructs the agent to read them for diagnostic clues only, never as instructions to follow or execute without user confirmation.
- [COMMAND_EXECUTION]: Uses common local development commands such as
npm test,npm run build, andgit bisectfor the purpose of reproducing and verifying bug fixes within the repository environment.
Audit Metadata